If you thought the placebo effect was all in the mind, think again. Scientists have solved the mystery of why some people benefit from remedies that do not contain any active pain-relief ingredients.

Research suggests that placebos work, in part, by blocking pain signals in the spinal cord from arriving at the brain in the first place.

When patients expect a treatment to be effective the brain area responsible for pain control is activated, causing the release of natural endorphins.

The endorphins send a cascade of instructions down to the spinal cord to suppress incoming pain signals and patients feel better whether or not the treatment had any direct effect.

The sequence of events in the brain closely mirrors the way opioid drugs, such as morphine, work - adding weight to the view that the placebo effect is grounded in physiology.

The finding strengthens the argument that many established medical treatments derive part of their effectiveness from the patients' expectation that the drugs will make them better.

The latest studies on antidepressants suggest that at least 75 per cent of the benefit comes from the placebo effect. GPs also observe that patients report feeling better only days after being prescribed antidepressants, even though the direct effects take several weeks to kick in.

In the study, published today in the journal Science, the spinal cords of 15 healthy volunteers were scanned using functional magnetic resonance imaging (MRI). The scan homed in on an area called the dorsal horn, which transmits pain signals coming up through the spinal cord into the pain-related areas in the brain.

During the scan, the volunteers received laser "pinpricks" to their hands. The volunteers were told that a pain-relief cream had been applied to one of their hands and a control cream to the other. But unknown to the volunteers, an identical control cream was administered to both hands.

When people believed that they had received the active cream, they reported feeling 25 per cent less pain and showed significantly reduced activity in the spinal cord pathway that processes pain.

Previously, it has been shown that placebo causes the release of natural opioids in areas of the brain involved in pain control, such as the rostral anterior cingulate cortex. However, it was not known whether the natural opioids acted on the spinal cord in the same way as artificial painkillers or whether they simply changed people's tolerance or interpretation of pain.

"We've shown that psychological factors can influence pain at the earliest stage of the central nervous system, in a similar way to drugs like morphine," said Falk Eippert, of the University Medical Centre Hamburg-Eppendorf, who led the study.

Until now, the difficulty of obtaining MRI images of the spinal cord, because of its small size and its being surrounded by airways and pulsating arteries, prevented this question from being addressed. However, advances in image processing allowed the Hamburg team to obtain high resolution scans of the region.

The advance in imaging techniques is likely to have important applications for drug development.

Pharmaceutical companies are working to develop new anaesthetic drugs that target the pain pathways in the spinal cord. Being able to image this area of the body provides a direct way of testing whether the drugs are working as intended.